do women in top management aﬀect firm performance? a ...correlations between ‘female-friendly’...

Munich Personal RePEc Archive

Do Women in Top Management Affect

Firm Performance? A Panel Study of

2500 Danish Firms

Smith, Nina and Smith, Valdemar and Verner, Mette

Aarhus University

2006

Online at https://mpra.ub.uni-muenchen.de/78715/

MPRA Paper No. 78715, posted 24 Apr 2017 10:03 UTC

February 2006

Do Women in Top Management Affect Firm Performance? A Panel Study of 2500 Danish Firms

Research paper

JEL codes: G38, J16; M14

Keywords: Firm performance, female CEOs, management diversity, gender diversity

Abstract

Purpose: This study examines the relationship between management diversity and firm performance in the case of women in top executive jobs and on boards of directors. Corporate governance literature argues that board diversity is potentially positively related to firm performance. This hypothesis is tested in the paper. Methodology: By the use of data for the 2500 largest Danish firms observed during the period 1993–2001 various statistical models for firm performance are specified and estimated. The main focus in the models is the estimated relationship between the proportion of women in top management (CEO’s and on boards of directors) and firm performance. Findings: The results show that the proportion of women in top management jobs tends to have positive effects on firm performance, even after controlling for numerous characteristics of the firm and direction of causality. The results show that the positive effects of women in top management strongly depend on the qualifications of female top managers. Originality: This paper provides solid statistical evidence of the effects of women in top management on firm performance. The use of a large sample and the panel nature of the data set make it possible to properly control for direction of causality and, furthermore, numerous firm and individual information are included to estimate genuine effects of women in top management. We wish to thank Anna Kossowska for competent assistance in collecting the data and computation work. We also thank two anonymous referees, Laura Rondi, CERIS-CNR, Torino, and other participants at the 6th Workshop on Corporate Governance and Investment held in Palma de Mallorca, February 2005 for many helpful comments.

2

1. Introduction

During the latest decade, there has been an increasing focus on the gender of top

executives and boards of directors of firms. The proportion of women reaching top positions

is still very low in most countries, though it has been increasing in for instance the US and in

some European countries. Some governments, like in Sweden and Norway, have even

introduced regulations of the gender composition of the boards of directors of private firms in

order to improve equal opportunities. In Norway, the government has decided that for large

Norwegian firms at least 40% of the members of the boards of directors must be women in

2005. This seems to have had a major impact on the recruitment practices for Norwegian

board members, see Hoel (2005). According to Hoel, the proportion of women in Norwegian

listed firms increased from about 6% in 2000 to 22% in 2005.

Parallel to this discussion, focus has been on good corporate governance in many

countries (see for instance for the US TIAA-CREF (2004) and for Denmark Nørby Johansen

et al. (2001)). One of the aspects of good corporate governance is management diversity, i.e.

a heterogenous composition of top managment. If it is actually the case that more women (or

minority groups) as top executives or members of boards of directors have a positive effect

on shareholder value and firm performance, this may be a strong argument for having more

women in top management.

In this study, we analyse whether female top executives and women on boards of

directors have any significant effect on firm performance measured by alternative

performance measures. The study examines the relationship between management diversity

and firm performance for the 2500 largest Danish firms observed during the period 1993–

2001. Management diversity is defined as the proportion of women among the highest

3

ranking CEOs in firms and on boards of directors. We estimate various panel data models of

firm performance and control for factors that are traditionally found to affect firm

performance e.g. firms’ age, size, sector, export orientation. We find that after controlling for

these observed factors, the proportion of women among top executives and on boards of

directors tends to have a significantly positive effect on firm performance. A large part of this

effect is attributed to the female managers with the best qualifications in terms of education,

and for the female board members it appears that the ones representing the staff have the

largest positive impact on firm performance. However, when controlling for unobserved

firm-specific factors, the effect often turns insignificant. This may reflect that until now very

few Danish firms have had women at the CEO level, and thus panel estimates of the

performance effects of female CEOs are determined with a large statistical uncertainty. An

alternative explanation may be, that the relatively few firms who hire women at the top level

of their organization are firms which are also doing well on a number of other unmeasured

characteristics (for instance good working conditions and work environment, a more

focussed recruitment policy etc.). Another crucial issue is the direction of causality (i.e. do

women on boards really affect firm performance or is it actually the case that better

performing firms are more likely to hire women?). Therefore, tests for causality between the

gender proportion on boards of directors and firm performance are performed. We find that

the positive relationship is due to board diversity affecting firm performance, not the

opposite.

2. Theoretical considerations and earlier findings

4

There are a number of arguments in favour of diversity of board members to be found

in the previous literature, see for instance Bantel and Jackson (1989) and Murray (1989).

Carter et al. (2003) list 5 positive arguments from a ‘business case perspective’ and also

discuss management diversity in a principal agent framework. Among the arguments pro

management diversity is that a more diverse board of directors (or executive board) is able to

make decisions based on the evaluation of more alternatives compared to a more

homogenous board. Women directors may have different experiences from their working life

and non-working life compared to men. They may have a better understanding than men of

some of the segments of the market place of the firm, and this may improve the creativity and

quality of the decision making process of the board, see Singh and Vinnicombe (2004). A

more gender diverse board may also improve the image of the firm and in this way have

positive effects on firm performance and shareholder value if the positive image has positive

effects on customers’ behaviour. Another argument for aiming at a more diverse composition

of board members is that if only male individuals are potential candidates for the boards, the

selection of board members will take place from only this selected distribution of

qualifications, and on average this implies a much lower quality than if the candidates are

selected among the best from the distribution of both men and women (or include minority

groups). Furthermore, women in top management positions may have positive effects on the

career development of women at lower levels within the organizations through mentor and

role models effects. Women at senior levels may affect positively the career aspirations of

younger women in lower positions, see Ely (1990), Burke and McKeen (1996), and Bell

(2005). This may increase firm productivity directly and more indirectly through a larger

pool of potential candidates for top positions in the firm.

5

However, there may also be arguments against management diversity. If a

heterogenous board produces more opinions and more critical questions, this may be time-

consuming and may not be as effective as a more homogenous board of directors, especially

if the firm is operating in a highly competitive environment where the ability to react quickly

to market shocks is an important issue. A culturally, ethnically or gender diverse board may

experience more conflicts, and even though the decisions may have a better quality in the

end, this may not balance the negative effects of a more slow decision-making process if the

market place of the firm demands quick responses, see Hambrick et al. (1996). Thus, based

on theory, the answer concerning the financial effects of management diversity and women

on boards is undetermined a priori.

Predictions from the previous empirical evidence are ambiguous. Most of the

empirical studies have been based on US data,1 and most of the studies include only the

largest firms. Shrader et al. (1997) analyse the 200 largest US firms and they are unable to

find any significantly positive relationship between the percentage of female board members

and firm performance (measured by return on assets, ROA, and return on equity, ROE). They

even find significantly negative relations in some cases. Kochan et al. (2003) also find no

positive relations between gender diversity in management and firm performance for US

companies. Contrary to these findings, Catalyst (2004) and Adler (2001) find positive

correlations between ‘female-friendly’ US Fortune 500 firms and the performance of these

firms.2 Carter et al. (2003) also find a significantly positive effect of the percentage of

1 Carter et al. (2003) give a survey of the empirical results for the US. 2 Catalyst (2004) defines ‘female friendly’ firms as the Fortune 500 firms which are ranked in the top 25% of the distribution with respect to proportion of women among 5% earners or in top management. The proportion is calculated as an average during the 5 year period 1996-2000. Average firm performance among upper quartile firms is compared to average firm performance among the least female friendly firms in the bottom quartile. Adler (2001) defined ‘female friendly’ firms as the top 25 Fortune 500 firms with the highest ‘score’ with respect to employing women as top managers during the period 1980-1998.

6

women and minorities on boards of directors and firm value after controlling for a number of

other factors which may affect firm value. The study by Carter et al. (2003) also controls for

the direction of causality by estimating an instrument variable model (IV-model), see below.

Only very few studies from outside the US exist on this topic. Du Rietz and

Henrekson (2000) analyse firm performance and women on boards for a sample of Swedish

firms. They find that if not controlling for firm size and sectors, firms with women on the

board seem to under-perform. However, when controlling for these factors, the under-

performance hypothesis could not be confirmed. A recent Norwegian study by Böhren and

Ström (2005) finds a significantly negative relationship between gender diversity (proportion

of women among board of directors) and firm performance (measured by Tobin’s Q) in

Norwegian listed non-financial firms observed during the period 1989-2002. For Denmark, a

study by Rose (2004) finds a negative, but insignificant relationship between the percentage

of women on the boards of directors and firm performance (measured by Tobin’s Q). The

study by Rose is based on cross section data for the 116 largest listed Danish firms. Thus, his

study is based on a quite small sub-sample of the firms included in the analysis presented in

this paper.

Thus, the conclusion from the previous empirical studies is ambiguous. Besides the

ambiguous theoretical predictions, the diverse empirical evidence may be due to different

estimation methods. In some studies, no controls for other factors are included. For instance

size and age of the firm (which are factors known to affect firm performance) may correlate

with the percentage of females on boards, and thus it may blur the picture if not controlled

for. Further, there may be a number of other unobserved factors which are important for firm

performance, but which will perhaps never be observable for the researcher. Therefore, panel

7

data where the same firms are observed in a number of years may give a more reliable

picture than cross-section studies based on only one year of observation.

A further problem with many of the existing studies is that the samples used are

typically only based on the largest (listed) firms for which it is possible to get reliable

information. Therefore, the results may not be representative for all firms in a given country.

Finally, it is important to control for the direction of causality. If well-performing firms

decide to employ more women (or minorities) because they decide on a more risky strategy

with respect to recruiting board members, the observed relationship between a gender diverse

board and firm performance will tend to become positive. If this is the case, causality may

run from performance to management diversity and not the reverse.

3. Data and methods

In this study, we aim at overcoming a number of the weaknesses by using a rather

unique data set based which is fairly large, compared to many earlier studies. Further, we

have information on the same firms during a a number of years and thus we are able to

control for unobserved heterogeneity among firms which may affect some of the earlier

estimates of the relationship between gender diversity and firm performance.

3.1. Sample description

The data set is an unbalanced panel of the 2500 largest Danish firms observed during

the period 1993-2001. Since Denmark is a small country and since there are few large firms

in Denmark, this means that our sample include fairly small firms. The average firm size in

the sample is 219 employees in 2001. The data set is based on register information from

Statistics Denmark who collects information for all Danish firms on a number of firm

8

characteristics for administrative purposes. Data includes extensive information on the firms

and the characteristics of the board members and thus allows us to use panel estimators and

control for causality. The information on firms is merged with individual information on the

employees of the firm, including information on background characteristics of the CEOs and

their spouses (spouse information is used in the statistical analysis to control for reverse

causality between firm performance and gender diversity). However, since Statistics

Denmark does not collect information on for instance board memberships, the information

from administrative registers has been merged with information from a private Danish data

register KOB (Købmandsstandens Oplysningsbureau). KOB collects information on

economic performance, board members of the firms etc. based on annual company reports to

the authorities. These reports are publicly available. KOB collects and harmonises this

information.

The sample is selected from the administrative registers as the 3000 largest Danish

firms, defined by gross turnover of the firms during each of the years 1993–2001. We

exclude companies with extreme values, defined as either a negative value of net capital or

an extreme relationship between firm’s revenue and employment in order to get rid of

holding companies etc. This means that the effective sample is reduced to about 2300-2500

firms for each of the years. For some of the variables, for instance membership of board of

directors, information is only available for the latest part of the period (1996–2001). Since

the sample is unbalanced, firms which are close to the cut-off criterion of being among the

3000 largest Danish firms may drop out of the sample in some years.3 The sample consists of

3 A potential problem using unbalanced panel data is that firms are entering and leaving the data set and that these changes are likely not to be random. Firm exits depend on performance, which potentially may depend on the gender composition of the board. In order to analyse this potential problem, we have calculated the sample proportion of women in top management for exit and entry firms and tested whether this proportion was higher

9

listed as well as non-listed firms. In Denmark, there are approximately 300 listed firms. The

selection criterium implies that most of these firms are selected into the sample. However,

most of the firms included in the data are not listed firms. Consequently, this data set is much

broader and more representative compared to the samples used in many other studies of

women in management.

3.2. Variable measurement

Four alternative measures on firm performance are available in the data. We use all

four alternative measures in the analyses in order to test the robustness of our results:

1. Gross profit/net sales 2. Contribution margin/net sales 3. Operating income /net assets 4. Net income after tax/net assets

Gross profit/net sales and contribution margin/net sales are both approximations of

firms’ markup. Gross profit is measured as net turnover (net sales) minus input expenses

(cost of good sold). Contribution margin is defined as net sales minus variable costs.

Consequently, both measures relate to firms’ basic activities. Operating income is defined as

the net result of the firm taking into account deductions and financial payments, but not

inclusive extraordinary revenues and expenses.

In Denmark, the management of private firms is organized as a two tier system. The

board of directors which is chosen by the stockholders typically consists of external board

members (except for staff members representing the employees of the firm, see below). The

chairman of the board (president) is also usually an external member, i.e. the CEOs are

for exit and entry firms than for firms who stayed in the sample during all years. We did not find any significant differences in the female proportion for entry and exit firms compared to firms who stayed in the sample. The results presented in Tables 3-9 do not change much when restricting the sample to a balanced panel which includes only about half as many observations as the unbalanced sample.

10

usually not formally members of the board of directors. In family owned firms, the

organization of the management is often different. Here, the top CEO (often member of the

owners’ families) may also be chairman of the board of directors. Further, it is very common

that the board of directors consists of mainly members of the owners’ families.

In order to measure gender diversity in management, we apply a number of alternative

measures of gender diversity in management. The most restrictive definition includes only

the proportion of women among the top CEOs in the firm.4 However, as a large proportion of

Danish firms have only one top CEO, we also introduce a broader definition of management

including vice-directors.

Further, we have information on the proportion of women on the boards of directors.

According to Danish law, a number of board members are selected among the staff in firms

with more than 35 employees. The number of staff representatives depends on the size of the

board. Traditionally, the proportion of women among the board members who represent the

staff is larger than among other board members. In some of the estimations, we distinguish

between the two types of board members.

In order to control for other factors, which may affect firm performance, a number of

other variables are included in the analysis. Firm size is measured by the number of

employees, i.e. performance is expected to be positively related to firm size, because larger

firms normally have more market power, Bain (1951), Smirlock et al. (1984). Younger firms

are expected to have smaller earnings than older ones, because they have less experience in

the market, are in a phase of building up their market position, and they normally have

relatively higher capital costs as compared to older firms, see e.g. Lipczinsky & Wilson

4 The measure which includes only top CEOs is defined as Disco 1, 12, 121 according to the occupational codes applied by Statistics Denmark, while the category ‘top CEOs plus vice-directors’ also includes Disco 122 and 123.

11

(2001). However, older firms may become more lax or be at a point on their product life

cycle with declining earnings. Thus, it may be argued that the effect of firm age is an inverse

U-shape. We also control for the export orientation of the firm. Firms that are heavily

engaged at export markets (high export/turnover rate) operate in potentially larger markets,

and this is expected to affect their profits positively. In line with other studies, we control for

potential effects on profit due to entry barriers for the concerned industry by using a measure

of the minimum efficient scale (MES) to the market (industry) size, where MES is

approximated by the first quartile firm’s turnover within the particular industry (measured at

the 4-digit level).5

3.3. Descriptive statistics

Table 1 shows mean statistics for the variables included in the data set. Note that due

to missing information, the number of observations varies for each variable. The reported

number of observations is the maximum number for the sample. The data is divided into two

groups: Firms with at least one female top CEO and firms without any female top CEOs.

(Table 1 about here)

From the table it is seen that firms having at least one female top CEO also have a

higher fraction of women in the lower ranks and also in the boards of directors. Thus, there

seems to be firms that are more “female-friendly” than others. Considering the means of the

four performance measures, there is a tendency of firms with female top CEOs doing worse

than firms with no female top CEOs in 2001. There is substantial variation in the proportion

of female managers across industries. Some industries (primary sector, energy and water) 5 MES is calculated by using the total sample of Danish firms which is approximately 25000 firms.

12

have no firms with a woman among their top CEOs or vice-directors, while other industries

are more ‘female-friendly’ like private service and retail, hotel and restaurants. This clearly

indicates substantial industry differences and that this is likely to affect the estimation results.

From Table 1 it appears that mainly older firms and firms with more employees tend

to have female managers. However, strong conclusions should not be drawn from these

simple descriptive statistics, since other characteristics are not controlled for.

Figure 1 shows the development during the period 1993–2001 in the proportion of

women among top CEOs and on boards of directors for all the firms included in this study. In

2001, 4.3% of the top CEOs in the largest Danish firms were women. When extending the

top management category to include vice-directors, this figure increases to 10.9%. The

proportion of women on boards of directors was 9.7% when including staff representatives

and 7.9% when excluding staff representatives.

(Figure 1 about here)

The female proportion among CEOs has been slightly increasing during the 1990s

from 2.5% in 1993 to 4.3% in 2001. When including vice-directors, the proportion of women

in management has almost doubled, from about 6% in 1993 to 11% in 2001. However, the

female proportion on boards of directors has declined. It decreased from about 12% in 1996

to less than 10% in 2001. The decline is mainly due to a relative reduction of female

members of boards of directors who are not representing the staff. One explanation may be

that due to the ethics of good corporate governance, still fewer family members are members

13

of boards of directors, and this may have reduced the proportion of women. Our data does

not allow us to test this hypothesis.

It is difficult to compare with other countries because of different definitions of top

management and variations in the samples selected. But the figures above seem to be

relatively low, see Table 2. The Danish figures based on the sample used in this study include

many relatively small firms. According to Table 1, the average firm size in the Danish

sample was 219 employees in 2001. If only the 113 largest firms in the sample are

considered, the proportion of females among top CEOs was slightly higher (5.9%) in 2005,

see Table 2. However, this is still much lower than in the US Fortune 500 firms where 10.2%

of the top CEOs were women in 2000. In Norway and Sweden, the proportion of women in

boards of directors has risen dramatically the latest years, clearly as a reaction to government

regulations in Norway, see Hoel (2005).


4. Results

4.1 Basic estimations

In general, the statistical model of firm performance can be written as,

(1) 1 2 3_ _it it it it itP X W CEO W boardβ β β ε= + + +

where i refers to the firm, and t is time. it

P is a performance measure. As described in the

data section, four performance measures are applied: Gross profit/net sales, Contribution

margin/net sales, Operating income /net assets and Net income after tax/net assets.

_it

W CEO and _it

W board are the key variables of this study, i.e. the proportion of women

in management, measured as (1) Top CEOs, (2) Top CEOs plus vice-directors, (3) Members

14

of boards of directors . Hence 2β and 3β are the parameters of primary interest. itX is a vector

of other explanatory variables typically assumed to affect firm performance, i.e. firm size

(number of employees), firm age (years since establishment, potential market size (measured

by the export-intensity of the firm))6 and an indicator of entry barriers (minimum efficient

scale). In addition, we add controls for industry and year of estimation in order to deal with

changing business conditions. εit is an error component, assumed to be Nid(0,σε2).

Since we have a panel of firms, we are able to control for time-constant unobserved

heterogeneity which may bias the results from cross-section studies if these unobserved

factors correlate with the proportion of female CEOs. Thus, we also present random and

fixed effects models of Pit. The panel version of equation (1) is:

(2) 1 2 3_ _ ( )it it it it i itP X W CEO W boardβ β β α ε= + + + +

where αi is the unobserved heterogeneity term, assumed to be firm-specific and time-

invariant. The random effects estimator is only valid if αi is uncorrelated with the explanatory

variables. We test the validity of the random effects estimator by a Hausman test. As shown

below, the Hausman test tends to reject the random effects estimator and thus the fixed

effects results are preferred.

The fixed effects estimator is consistent, but it does not give any estimates on

variables which are time constant. Further, for variables with a small variation, the estimates

are imprecise (has a large variation) and therefore results tend to become insignificant due to

small variation across time. The key variable in this study, the proportion of women in

management, suffers from this problem. Therefore, we first present estimation results from

pooled estimations of (1), see Table 3, where selected key results of pooled ordinary least

6 The estimation forms allow for non-linearity by including squared expressions for firm size and age, see the theoretical discussion above.

15

squares (OLS) regressions based on the sample of firms observed during the period 1993 (or

1996 for board of directors estimations) to 2002 are presented.7 For each of the four

performance measures, we have estimated four alternative models, including a number of

explanatory control variables as described above (firm size, firm age etc.) plus the

proportions of women in top management and the share of women on boards of directors.8 In

the first row of Table 3, the effect of a female top CEO is shown, the second row of results

shows the coefficients of the female proportion among top CEOs and vice-directors, the third

row shows the coefficients including the variable proportion of women on boards of

directors.


The first group of results concerns the effect of female top CEOs. In general, the

estimated coefficients are positive, but except for column 1 (gross profit/net sales) no

significant effects are found. However, extending the definition of top management to

include vice-directors, the estimated coefficients turn significant for three out of four firm

performance measures. Thus, when controlling for firm size, sector, age of firm etc., we find

that there is a positive performance effect of female CEOs for Danish firms.

Turning to the female representation on boards of directors, the results are more

mixed. When including a variable measuring the proportion of women among all board

7 Due to space considerations we do not show the full estimation results for all the models in this paper. The full estimation results are available from the authors upon request. 8 For different reasons, we do not include both W_CEO and W_board in the same regressions but estimate their effects in separate regressions. Firstly, W_board is only observed for a sub-period, i.e. the years 1996-2001. Second, we are not able to instrument both variables when controlling for reverse causality, and thirdly, the two variables are correlated, i.e. we have problems of multicollinarity, see correlations coefficients in Appendix. Thus, the ‘total effect’ of women in top management is not the sum of the two estimated effects.

16

members, there is only one positive and significant coefficient found (contribution

margin/net sales) while the coefficient for some performance measures is negative, though

insignificant.

In order to control for unobserved time-constant heterogeneity, which may affect firm

performance, we have also estimated alternative panel data models. Table 4 shows the results

from fixed effects and random effects estimation the model. The reported Hausman tests

reject the random effects model in all cases; i.e. the random effects parameters are likely to

be biased from correlation between the firm-specific effects and the explanatory variables.

Therefore, the preferred panel data specification is the fixed effect model. As expected, the

level of significance drops considerably when moving from the pooled OLS to the fixed

effect estimation. This may be the result of a very small variation in the explanatory

variables, including the key variables on female proportions in top management or

measurement errors which reinforce the problems of getting significant coefficients. An

interesting result is the significantly negative coefficient for the variable representing female

proportion among board of directors in Columns 1-2. The interpretation of this result is

discussed further below.


4.2 Direction of causality.

The direction of causality between firm performance and the proportion of women in

management has been widely discussed. Thus, according to this discussion some firms may

be observed to have a high proportion of female CEOs because these firms are currently

17

doing well and may be able ‘to take the risk’ of employing a female CEO. If this is the case,

the direction of causality is the reverse in relations (1)–(2), i.e. profitable firms have

relatively more female managers than less profitable firms. In order to deal with problems of

potential endogeneity in equations (1)–(2), we estimate the models by the use of an

instrument variable approach (IV). The proportion of women in management in the firm is

then estimated by the following equation:

(3) 1 2_ it it it itW CEO X Zα α ν= + +

where Xit is a vector of firm characteristics inclusive firm performance, itZ are the

instruments, i.e. factors affecting the proportion of women in management, which at the same

time do not affect firm performance, and νit is an error component, assumed to be Nid(0,σν2).

Since _it

W CEO is a proportion, 0≤ _it

W CEO ≤1, a linear specification of (2) is inappropriate.

Therefore, (3) is estimated by a tobit estimator which takes into account the upper and lower

limits of the dependent variable, _it

W CEO .

It is a well-known problem to find valid instruments, i.e. Z-variables. This is also

difficult in this context because there are no obvious variables included in our data set which

clearly explain female proportion in top management without being suspected to influence

the observed firm performance. We have tested more variables related to firm characteristics,

but these variables did not pass a test of being a valid instrument based on the method

described in Bound et al. (1995).9 Instead we use as an instrument the average length of

education of the spouses of the other CEOs in the firm. Our hypothesis is that CEOs who are

married to well-educated spouses (in most cases these are wives) are supposed to be more

9In the test proposed by Bound et al. (1995) it is analysed whether the instrument significantly affects the proportion of female managers in a firm, but has no direct significant effect on firm performance.

18

positive and have a less traditional view on the competences of female CEOs, implying that

they are more willing to hire and accept a woman in their own firm as compared to other

CEOs who are married to lower educated spouses. Furthermore, the education of the spouse

is assumed not to affect the firm performance. The test by Bound et al. (1995) cannot reject

that the instrument is valid. However, we are only able to apply an IV-estimator in the

models analysing the effects of female CEOs. We do not have individual information on

board members since the information on board members does not stem from Statistic

Denmark’s administrative registers. Thus, we do not have valid instruments for the

regressions where _it

W board enters.

In Table 5, the estimates for panel versions of the IV-models are presented. As the

previous results show that the fixed effect model is strictly preferred to the random effects

model, only models in which the second step is estimated as pooled OLS and fixed effects

estimation are presented.


The results in Table 5 confirm the results presented previously. The estimates from

IV-pooled OLS show positive, significant relationships between female CEOs and

comparing to the results of Table 3 (pooled OLS, no IV) the introduction of IV to the pooled

OLS generates larger and more significant positive effects of women in management.

However, the results from IV-fixed effects models are again insignificant. Thus, our results

document, that when controlling for causality, there is still a positive effect of female top

CEOs and vice-directors. As in Table 3, the size of the effect is larger for the broader group

19

including vice-directors comparing to the effect observed for only top CEOs. One

explanation of this evidence may be that the relatively large increase in Denmark during the

period 1993-2001 in female proportion of vice-directors reflects that still more highly

educated women are now potential candidates for being promoted into top-management due

to their qualifications and professional skills. Older female top CEOs may to a larger extent

have been selected due to family ties to the owners’ families. We will pursue this hypothesis

below.

Though we are able to reject a hypothesis of reverse causality in our study, it is

important to stress that the interpretation of our results is not as straightforward as it might

seem in first instance. Since we find significantly positive coefficients in the pooled OLS

estimations, but insignificant panel estimates (fixed or random effects), this may reflect that

the firms who have succeeded in hiring female top managers are the firms with the most

ambitious/progressive/active characteristics in general. If there exist unobserved

characteristics of this type in the firm, which are (almost) time constant during our

observation period, this may at the same time explain that more women are hired in top

management. For instance firms with an explicit management diversity policy or, in general,

a more ambitious recruitment policy may also in other respects have characteristics not

observed (good working environment, high degree of team spirit etc.) which explain firm

performance and profits.

4.3. Qualification effects

In the previous analysis, we have not looked at the qualifications of female managers,

i.e. the estimated effects are average effects for all women in top management. However, the

estimated effects potentially conceal variations in effects of qualifications of the female

20

managers. The hypothesis is that the positive effect of women in top management is larger,

the more qualified the female managers are. If many of the women who are observed as top

managers in our sample were mainly selected back in time because they were for instance

widows or daughters of the owner(s) and not because of their formal qualifications relevant

for a position as a top manager, the estimated coefficients above may not give an adequate

picture of the potential for a future policy of management diversity and hiring (formally

qualified) women as top managers.

In Table 6, the educational distribution of the CEOs is presented. In 1993, 74% of all

female top CEOs in the sample did not have a formal higher education, while this was only

the case for 59% of the male top CEOs. These figures change considerably during the 1990s,

and in 2001 only 46% of female and 44% of male top CEOs had no higher education. The

proportion of female top CEOs holding a masters degree (long higher education) almost

doubled, from 17% in 1993 to 32% in 2001, while the same figures for male top CEOs were

30% and 36%, respectively. Thus, the educational level of female top CEOs has increased

much faster than that of male CEOs, and the educational gap has narrowed considerably

during the period. When including vice-directors, the picture changes. Female CEOs at the

level just below the top CEO are on average as well educated as their male peers. Since the

broader category of CEOs including vice-directors is younger than the group of top CEOs on

average, this indicates a development over the period 1993-2001 where Danish female CEOs

have become still more qualified and have almost closed the qualification gap to their male

colleagues.


21

In order to test whether the differences in formal competences matter for the

estimated performance effects of women in top management, we split the variable W_CEO

into three variables which measure the proportion of females among the group of CEOs at

different educational levels

(4) _ _ _ & _it it it it

W CEO W long W M S W EDU= + +

where W_longit is the proportion of women among the group of CEOs having at least an

education at the master level (long higher education), W_M&Sit is the proportion of women

among the group of CEOs having an education at the BA-level or slightly shorter (short or

medium higher education), W_EDUit is the proportion of women among the group of CEOs

having no education or vocational education. Consequently, we estimate the form:

(1a) 1 2 3 4_ _ & _it it it it it itP X W long W M S W Otherβ β β β ε= + + + +

Since there are few females in the group of top CEOs, we only estimate (4) for the

group of CEOs including vice-directors. The results of these estimations (see Table 7) are

striking, in the sense that they show that a large part of the performance gain from female

managers can be attributed to the highest educated women for whom the estimated

performance effects tend to be much larger than for women with a short or medium higher

education (except for Column 2, Profit on primary operations). For women with no higher

education, i.e. women who are either unskilled or have a vocational education, the

performance effect is larger than for women with a short or medium higher education.10

10 This result may be explained by the fact that short and medium long theoretical educations in Denmark to a large extent are educations related to service and education jobs in the public sector (nurses, teachers, care workers etc.). These educations are typical ‘female educations’ which may be very different from educations typical for a career in the private sector. One hypothesis might be that women with these types of educations who end up as CEOs often have family ties to the owners. We are not able to test this hypothesis because we lack information on family ties.

22


Among the board members, similar mechanisms may play a role. Only limited

information about the board members is available. However, we do have information on the

board member status of the individual, i.e. whether the board member is a staff representative

or not. From Table 8 it is seen that the proportion of females among staff board members is

substantially higher than among other board members, and furthermore it appears that,

contrary to the general negative trend of women in the Danish boards, the female share

among the staff members has increased slightly over the observation period.


If we assume that women who are selected into the board of directors are as qualified

as their male colleagues who are staff representatives on the board of directors, and if a

proportion of female non-staff board members are selected because they have family ties to

the owners and not because of their educational qualifications, we will expect that a potential

positive effect of females on boards is larger for female staff members than for female non-

staff members.11

In Table 9, this hypothesis is tested. The first row of estimates corresponds to the

estimates for the female board member effects presented in Table 3 where only one small

positive significant effect was found. When we split the female board member effects, i.e.

11 For the 100 largest Danish firms more than two thirds of all female supervisory board members (excl. staff representatives) have family ties to the owners.

23

estimates two separate performance effects for the female proportion on the board, an

interesting picture emerges: 7 out of 8 estimates are now significant and for the staff

members of the board the four coefficients are positive and very significant. On the other

hand, for female non-staff members of the board the effects are significantly negative and

rather large in two out of four cases.


Thus, female non-staff members of boards of directors seem to have a less positive or

even negative effect on firm performance compared to staff members. This finding may

indicate that firms who employ family members on their board of directors seem to be less

successful with respect to performance compared to firms who do not.12

5. Conclusion

The purpose of this paper is to evaluate the influence of the proportion of women in

management on firm performance. There is a theoretical as well as an empirical motivation

for dealing with this issue of corporate governance. In the theoretical section of the paper, we

argue that board diversity affects the performance of the firm. However, according to the

existing theory the influence can be positive as well as negative. The empirical motivation

comes from the increasing focus on the gender composition of top executives and boards of

directors of firms. The proportion of women who reach top positions in the business sector is

still very low in most countries, though it has been increasing in some countries. If it can be

12 This corresponds to the results of Bennedsen et al. (2005) who find for a sample of Danish firms that family successions in CEO positions have a negative impact on performance compared to non-family successions.

24

shown statistically that more female top executives (or other minority groups) or women on

boards of directors have positive effects on firm performance, this may be a strong argument

for having more women on boards.

Using a sample of the 2500 largest Danish firms over the period 1993-2001, we

analyse empirically whether the proportion of female top CEOs or members of boards of

directors really affects firm performance. The conclusion is ambiguous and depends both on

the measure of performance and the measure of the proportion of women in management.

The effect on firm performance of a higher fraction of female top CEOs varies from none to

positive. Performance measures which approximate the mark-up, e.g. gross profit are

affected more positively and more significantly than the other performance measures, e.g. net

income after taxes. Furthermore, the results show that the positive performance effects are

mainly related to female managers with a university degree while female CEOs who do not

hold a university degree have a much smaller or insignificant effect on firm performance.

Next, female members of boards of directors elected by the staff seem to have

positive effects on firm performance. However, this positive effect does not carry over to

other female board members, where the effect is negative - a result, which may be explained

by the fact that a significant part of the women on boards have family ties to the owners.

However, we are not able in this study to identify whether the board members have family

ties or not. An important topic for future research would be to identify the performance

effects of female - as well as male board members - who have family ties to the owners.

The question concerning getting more women on boards and in top executive jobs is a

highly debated issue in many countries. In Norway, there have already been political

initiatives regulating the proportion of women among board members. Our results are to

25

some extent supporting the view that a more gender diversity in top management positions

would improve the financial performance of Danish firms. However, since our results also

indicate that qualifications are important, it is important that there is a sufficient potential

pool of qualified women who can fill the positions as board members or who might be

recruited as top CEOs. Therefore, the main implication from our study is the importance of

attracting and recruiting more women into the higher ranking positions in firms and thus

increasing the number women who are qualified to be selected into boards of directors or as

top CEOs. It may have negative boomerang effects in the longer run if for instance very rigid

quotas are imposed on the composition of board members.

References

Adler, R.D. (2001), Women in the Executive Suite Correlate to High Profits, Harvard Business Review.

Bain J.S. (1951), Relation of profit rate to industry concentration: American manufacturing, 1936-1940, Quarterly Journal of Economics, 65, 293-324. Bantel, K.A. and S.E. Jackson (1989), Top Management and Innovations in Banking: Does the composition of the Top Team Make a Difference?, Strategic Management Journal, vol. 10, (Special Issue), 107-124.

Bell, L. A. (2005), Women-Led Firms and the Gender Gap in Top Executive Jobs, IZA discussion paper 1689, IZA, Bonn.

Bennedsen, M., K. Nielsen, F Pérez-González and D. Wolfenzon (2005): Inside the Family Firm: The Role of Families in Succession Decisions and Performance, Discussion Paper 2005-14, Centre for Economic and Business Research, Copenhagen Business School. Böhren, Ö and R. Ö. Ström (2005), Aligned, informed, and decesive: Characteristics of value-creating boards, Working Paper, Norwegian School of Management BI, September 2005, Oslo.

Bound, J., Jaeger D.A., and R. Baker (1995), Problems with instrumental variables estimation when the correlation between the instruments and the endogenous explanatory variable is weak. Journal of American Statistical Association vol. 90 (430), 443-450.

26

Burke R. J. and C. A. McKeen (1996), Do Women at the Top Make a Difference? Gender Proportions and Experiences of Managerial and Professional Women, Human Relations, vol. 49 (8), 1093-1104.

Carter, D.A., B.J. Simkins and W.G. Simpson (2003), Corporate Governance, Board Diversity, and Firm Value, The Financial Review, vol 38, 33-53. Catalyst (2003), Women in US Corporate Leadership: 2003’, www.catalystwomen.org Catalyst (2004), The Bottom Line: Connecting Corporate Performance and Gender Diversity, www.catalystwomen.org

Du Rietz, A and M. Henrekson (2000), Testing the Female Underperformance Hypothesis, Small Business Economics, vol 14(1), 1-10. Ely, R. (1990), The Role of Men in relationships among professional women, Academy of Management Best Paper Proceedings, 1990, 364-368. Hambrick, D.C., T.S. Cho and M.J. Chen (1996), The influence of top management team heterogeneity on firms’ competitive moves, Administrative Science Quarterly, vol. 41, 659-684. Henrekson, M. (2004), How to improve equal opportunities in Swedish business sector (in Swedish), SNS forlag, Stockholm. Hoel, M. (2004, 2005), Women on boards and in top executive jobs in Norwegian business sector. A description of the largest corporations 2004 (in Norwegian), CCD/Ledelse Likestilling Mangfold, Oslo.

Kochan, T., K. Bezrukova, R. Ely, S. Jackson, A. Joshi, K. Jehn, J. Leonard, D. Levine and D. Thomas (2003), The Effects of Diversity on Business Performance: Report of the Diversity Network, Human Resource Management, vol 42 (1), 3-21. Likestillingssenteret (2001), Equal opportunities barometer (in Norwegian) 2001, Oslo. Lipczinsky, J. & J. Wilson (2001): Industrial organisation – An Analysis of Competive Markets, Prentice Hall, 2001. Murray, A.I. (1989), Top Management Group Performance and Firm Performance, Strategic Management Journal, Vol 10 (Special Issue, Summer 1989), 125-141. Nørby-Johansen, L., J. Lindegaard, W. Schmidt and M. Øvlisen (2001), The recommendations from the Nørby Committee concerning good corporate governance in Denmark (in Danish), Erhvervs- og Selskabsstyrelsen, Copenhagen.

27

Rose, C. (2004), The composition of boards and financial performance in Danish listed firms – are the recommendations from the Nørby Report beneficial for the shareholders? (in Danish) WP 2004-2, Institut for Finansiering, Handelshøjskolen i København. Shrader, C.B., V.B. Blackburn and P. Iles (1997), Women in Management and firm financial performance: an explorative study, Journal of Managerial Issues, fall 1997, vol. 9 (3), 355-372. Singh, V. and S. Vinnicombe (2004), Why so few women directors in top UK boardrooms? Evidence and theoretical explanations, Corporate Governance – An International Review, vol. 12 (4), 479-488. Smirlock, M., Gilligan T.W. & W. Marshall (1984), Tobins q and the structure performance relationship, American Economic Review, 74, 1051-60. TIAA-CREF (2004), Policy Statement on Corporate Governance: http://www.tiaa-cref.org/pubs/html/governance_policy/board_directors.html

28

Table 1. Sample means and standard deviations, 1993 and 2001. 2001 1993

Firms with

at least one

female top

CEO

Firms with

no female

top CEO

Firms with at

least one

female top

CEO

Firms with

no female

top CEO

Female Proportion: CEOs 0.499

(0.300) 0.567

(0.293)

CEOs and vice-directors 0.292 (0.204)

0.063 (0.154)

0.357 (0.269)

0.036 (0.114)

Board of directors, excl. staff 0.139 (0.184)

0.067 (0.138)

Board of directors, staff only 0.133 (0.251)

0.061 (0.180)

Firm Performance: Gross profit/net sales 0.353

(0.189) 0.305

(0.213) 0.398

(0.200) 0.325

(0.208) Contribution margin/net sales 0.036

(0.101) 0.045

(0.106) 0.042

(0.058) 0.045

(0.078) Operating income /net assets 0.210

(0.433) 0.234

(0.601) 0.707

(1.150) Net income after tax/net assets 0.151

(0.345) 0.160

(0.357) 0.178

(0.459) 0.176

(0.355) Sector(indicator variables, 0/1): Primary 0.0151

(0.122) 0.015

(0.122) Manufacturing 0.343

(0.477) 0.300

(0.459) 0.348

(0.481) 0.321

(0.467) Energy and water 0.021

(0.142) 0.006

(0.078) Building and construction 0.051

(0.220) 0.045

(0.208) 0.055

(0.227) Retail, hotel and restaurants 0.374

(0.486) 0.402

(0.490) 0.391

(0.493) 0.384

(0.486) Transportation, telecommunication etc. 0.040

(0.198) 0.050

(0.219) 0.043

(0.206) 0.044

(0.205) Private service 0.192

(0.396) 0.166

(0.373) 0.217

(0.417) 0.175

(0.380) Age of firm (year) 40.040

(37.482) 32.693

(32.981) 43.690

(25.886) 38.415

(32.802) Number of employees 626.061

(1644.250) 201.498

(609.022) 352.175

(684.714) 154.332

(427.710) Export(indicator variables, 0/1):: Low (less than 10% of turnover) 0.242

(0.431) 0.324

(0.468) 0.413

(0.498) 0.555

(0.497) High (more than 50% of turnover) 0.737

(0.442) 0.659

(0.474) 0.261

(0.444) 0.203

(0.402) Min. efficient scale relative to market size (MES = quartile turnover in industry, 4-digit)

8.433 (1.334)

8.566 (1.351)

7.873 (0.914)

7.925 (0.962)

Number of firms 99 2,380 46 2,108 The number of observations included in the last row is the maximum number of firms in the year concerned. For some of the variables there may be fewer observations than this number. Firms with extreme values for the performance variables have been excluded from the sample.

29

Figure 1. Proportion of women in top management and on boards of directors, 1993-2001.

0

2

4

6

8

10

12

14

1993 1994 1995 1996 1997 1998 1999 2000 2001

%

CEOs

CEOs plus vice-directors

Board of dir., incl. staff

Board of dir., excl. staff

30

Table 2. Proportion of women in top management (CEO) and on boards of directors, selected countries.

Country

CEO

Board of directors

US, Fortune 500 firms, 2000 (CEO)/2002 (board of

directors )

10.2%1) 13.6%

Sweden, 2002, large firms (sales > SEK 50 million)

Sweden 2005 (178 largest firms)

5.2%

15.0%

12.0%

18.7%

Norway, 2001, firms > 250 employees (CEO) and

listed firms (board of directors )

Norway, 2005 (97 largest firms)

4.5%

12.4%

6%

21.6%

Denmark, 2001, 2500 largest firms (sales), this study

Denmark, 2005 (113 largest firms)

4.3%

5.9%

9.7% (incl. staff)

7.9% (excl. staff)

11.7%

1) A broader definition of top CEO is used than the one used for Denmark. 2) Weighted average calculated on the basis of 380 and 109 firms who had, respectively had not, answered a survey on board membership. Information based on Hoel (2004, p. 13). Source: US, Catalyst (2003, 2004), Sweden, Henrekson (2004) and Hoel (2005), Norway, Likestillingssenteret (2001) and Hoel (2004, 2005), Denmark, Hoel (2005).

31

Table 3. Estimated coefficients of alternative variables reflecting proportion of women in

management (β2 and β3). Pooled OLS, 1993-20011).

Dependent variable: Firm performance

Gross profit/net sales

Contribution margin/net sales

Operating income /net assets

Net income after tax/net

assets Top CEOs 1993-2001

0.063* (0.012)

0.162

18,862

0.006 (0.004)

0.034 18,862

0.051 (0.034)

0.022

14,554

0.032 (0.022)

0.019 18,862

Top CEOs and Vice-directors 1993-2001

0.094* (0.009)

0.170

18,862

0.003 (0.003)

0.034 18,862

0.092* (0.027)

0.024

14,554

0.072* (0.016)

0.020 18,862

Board of directors 1996-2001

0.011 (0.011)

0.160

12,085

0.012* (0.004)

0.034 12,085

-0.051 (0.031)

0.022

12,080

-0.032 (0.021)

0.018 12,085

* Denotes that the estimated parameter is statistically different from 0 at the 5% level of significance. 1) Values in parentheses are standard errors of the estimated parameters, Adj R2 is reported in row 3, and row 4 includes the number of observations in each regression analysis.

32

Table 4. Estimated coefficients of alternative variables reflecting proportion of women in

management (β2 and β3). Fixed- and random effect estimations, 1993-20011).





Net income after tax/net assets

FE RE FE RE FE RE FE RE

Top CEOs 1993-2001

0.020* (0.006)

0.036 18,862

0.023* (0.006)

0.029

18,862 1417.5

-0.002 (0.004)

0.017 18,862

-0.001 (0.004)

0.015 18,862 97.2

-0.022 (0.039)

0.023

14,554

0.005 (0.035)

0.022 14,554 45.6

-0.026 (0.024)

0.014 18,862

-0.005 (0.022)

0.013 18,862 68.5


-0.002 (0.005)

0.036 18,862

0.005 (0.005)

0.028

18,862 2677.1

-0.004 (0.003)

0.017 18,862

-0.003 (0.003)

0.015 18,862 100.4

-0.019 (0.034)

0.023

14,554

0.022 (0.023)

0.021 14,554 48.2

-0.013 (0.020)

0.013 18,862

0.017 (0.017)

0.012 18,862 75.1

Board of directors, all 1996-2001

-0.029* (0.010)

0.062 12,085

-0.024* (0.009)

0.050

12,085 565.4

0.002 (0.006)

0.018 12,085

0.003 (0.005)

0.015 12,085 60.0

0.015 (0.053)

0.021

12,080

-0.017 (0.039)

0.019 12,080 39.1

-0.009 (0.037)

0.017 12,085

-0.014 (0.026)

0.015 12,085 39.7

* Denotes that the estimated parameter is statistically different from 0 at the 5% level of significance. 1) Values in brackets are standard errors of the estimated parameters, Adj. R2 is reported in row 3, and row 4 includes the number of observations in each regression analysis. The values reported in row 5 are Hausman Chi-square test for systematic differences in estimated coefficients.

33

Table 5. Estimated coefficients of alternative variables reflecting proportion of women in management (β2). IV-estimations, 2nd step: pooled OLS and fixed effect estimations, 1993-2001. Instrument: mean educational level of managers’ wives1).



Contribution

margin/net sales Operating income

/net assets Net income after

tax/net assets Pooled OLS

FE Pooled OLS

FE Pooled OLS

FE Pooled OLS

FE

Top CEOs 1993-2001

0.088* (0.008)

0.157 17105

-0.005 (0.006)

0.038 17105

0.013* (0.003)

0.035 17105

-0.003 (0.003)

0.019 17105

0.074* (0.023)

0.021 13132

-0.051 (0.035)

0.024 13132

0.043* (0 .014)

0.017 17105

-0.043 (0.023)

0.014 17105


0.365* (0.032)

0.157 17105

-0.019 (0.024)

0.038 17105

0.053* (0.012)

0.035 17105

-0.014 (0.015)

0.017 17105

0.306* (0.095)

0.021 13132

-0.210 (0.143)

0.024 13132

0.177* (0.059)

0.017 17105

-0.178 (0.094)

0.014 17105

* Denotes that the estimated parameter is statistically different from 0 at the 5% level of significance. 1) Values in parentheses are standard errors of the estimated parameters, Adj R2 is reported in row 3 and number of observations in row 4.

34

Table 6. Educational level among female and male CEOs. 1993 and 2001.1) 1993 2001 1993 2001 Top CEOs Top CEOs and Vice-directors

Women Men Women Men Women Men Women Men

Long higher education (master)

0.174 0.298 0.315 0.358 0.239 0.228 0.289 0.309

Medium or short higher education (bachelor)

0.087 0.113 0.226 0.206 0.094 0.049 0.159 0.136

No higher education .

0.739 0.589 0.459 0.436 0.667 0.723 0.552 0.556

All 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

No. of obs. 46 1020 99 1142 345 1322 641 1408

1) Long higher education is defined as 8 years or more after compulsory schooling. This typically means a masters degree from a university. Medium or short higher education is 6-7 years of higher education after compulsory schooling (bachelor or others), while no higher education is either vocational training or no other completed qualifying education.

35

Table 7. Estimated coefficients on effects of proportion of women in management, top CEOs

and vice-directors (β2). Pooled OLS, 1993-20011,2).

Dependent variable: Firm performance Proportion of

female CEO

with:





Long higher education (master)

0.283* (0.028)

0.009

(0.010)

0.148

(0.085)

0.121* (0.053)

Medium or short higher education (bachelor etc.)

0.014* (0.006)

0.000

(0.002)

0.029

(0.019)

0.017

(0.012)

No higher education

-0.026* (0.003)

0.003

(0.013)

0.069* (0.010)

0.045* (0.006)

Adj. R2 Nr. of observations

0.168 18862

0.036 18862

0.026 14554

0.022 18862

* Denotes that the estimated parameter is statistically different from 0 at the 5% level of significance. 1) See note 1 in Table 6. 2) Values in parentheses are standard errors of the estimated parameters.

36

Table 8. Proportion of women among members of boards of directors including and

excluding staff members. 1996 and 2001

1996 2001 Board of directors, all

0.117 0.099

Board of directors, excl. staff

0.101 0.080

Board of directors, staff only

0.194 0.207

37

Table 9. Estimated coefficients of effects of proportion of women among members of boards

of directors (β3). Pooled OLS, 1996-20011).





Net income after tax/net

assets Board of directors, all 1996-2001

0.011 (0.011) 0.160

12,085

0.012* (0.004) 0.034 12,085

-0.051 (0.031) 0.022

12,080

-0.032 (0.021) 0.018 12,085

Board of directors, excl. staff 1996-2001

-0.009 (0.011) 0.171

12,085

0.007 (0.005) 0.038 12,085

-0.079* (0.031) 0.029

12,080

-0.061** (0.021) 0.026 12,085

Board of directors, staff only 1996-2001

0.038* (0.010) 0.171

12,085

0.014* (0.004) 0.038 12,085

0.065* (0.028) 0.029

12,085

0.063* (0.019) 0.026 12,085

* Denotes that the estimated parameter is statistically different from 0 at the 5% level of significance. 1) Values in parentheses are standard errors of the estimated parameters, Adj R2 is reported in row 3 and number of observations in row 4.

38

Appendix: Pearsons correlation coefficients for key variables, 2001.





Proportion.of women among top CEOs

Proportion of. women among top CEOs and vice-directors

Proportion of women in boards of directors


1.0000 0.3611 0.0341 0.0569 0.0533 0.1110 -0.0313


0.3612 1.0000 0.2921 0.3656 -0.0151 0.0148 -0.0264


0.0341 0.2921 1.0000 0.7602 -0.0002 -0.0152 -0.0154


0.0569 0.3656 0.7602 1.0000 0.0092 -0.0153 -0.0002

Proportion of women among top CEOs

0.0533 -0.0152 -0.0002 0.0092 1.0000 0.4570 0.1407

Proportion of women among top CEOs and vice-directors

0.1101 0.0148 -0.0152 -0.0133 0.4570 1.0000 0.0599

Proportion of women among boards of directors

-0.0313 -0.0264 -0.0154 -0.0002 0.1407 0.0599 1.0000

do women in top management aﬀect firm performance? a ...correlations between ‘female-friendly’...

Documents